someone smarter than Dave Cutler
in [Device Drivers]
|
From: Peter Flass on 7 Apr 2010 08:05 Morten Reistad wrote: > In article <jfudndz3ZJOCTinWnZ2dnUVZ8uKdnZ2d(a)eclipse.net.uk>, > Dave <g8mqw(a)yahoo.com> wrote: >> "Patrick Scheible" <kkt(a)zipcon.net> wrote in message >> news:w9zfx3fktul.fsf(a)zipcon.net... > >>> IBMs were leased. Would IBM continue to support a computer that had >>> some academics' experimental hardware hooked up to it? Could new and >>> experimental device drivers be added to IBM's OS? These might be as >>> important as the machine's architecture. >> Well post anti-trust they had too. And I know certainly in the UK, we had >> lots of weird kit hooked into various University mainframes using a wide >> variety of interfaces. The oldest I remember NUNET/NUMAC (I think) used >> PDP/11s acting as IBM2708/3708 concentrators. There were "Browns Boxes" for >> X.25 and some how Cambridge Ring got connected in at Leeds but I think that >> was actually an Amdahl at that time... > > This was how mainframe networks were built. A cobbled-together mess > of protocols, none from IBM. Philips had some gear that was very > popular in banking/insurance networks in the mid 1970's, using > datex (circuit switched 1200 bps) to access a remote network, and > local end stations with local intelligence and some spooling. > > Similar stuff was used for travel agencies; but they used multidrop > bisync; but still with some local intelligence. And the transaction > screens became a nightmare of cryptical commands to get it all into > one transaction. > > None of this had a chance to scale beyond the corporate "network" > it was deployed in. I saw a merger between two banks up close, where > they really tried to integrate. No such luck. They had to replace > so they only had one network. > > Stats from internet exchange points still notice the encapsulated > bisync, x25, frame relay etc. in the protocol fields. These legacy > bits used 3% of the ix bandwidth as late as 2005. > It's especially sad seeing the difference between the 360 architecture, that was so well documented up-front, and SNA, that seeems to be whatever could be made to work at the time. It's like two different companies.
From: Peter Flass on 7 Apr 2010 17:50 Charles Richmond wrote: > Patrick Scheible wrote: >> jmfbahciv <jmfbahciv(a)aol> writes: >> >>> Mark Crispin wrote: >>>> On Tue, 30 Mar 2010, Jim Stewart posted: >>>>> Anyone that takes the time to leaf through some >>>>> Datamation magazines of that era would be lucky >>>>> to find any reference to PDP-10's. >>>> Using Datamation as an historical reference is like using the >>>> National Enquirier. >>> ROTFLMAO. Much better answer than the one I just wrote. >> >> The National Enquirer just makes stuff up for (dubious) entertainment >> value. Datamation didn't. Datamation just concentrated on the >> corporate data processing market, where IBM did indeed dominate. But >> the academic and research markets were a lot more fun and interesting, >> and they were dominated by PDP-10s. >> >> -- Patrick > > Didn't Xerox Data Systems (nee Scientific Data Systems) have any sort of > impact on academic and research markets??? ISTM that Xerox fumbled this > market too, but there were a lot of systems out there before Xerox gave up. > From what I know, I wouldn't say "a lot", although they were certainly there (or "in that space" as marketdroids would say today). They were a little late to the party and, although I'd rate their systems as "good" or "very good", I believe they didn't give most people enough reason to change. I'd say most of the existing software was IBM (PL/C, WATFOR, etc.) or DEC (fill in your choices here...)
From: Dave on 10 Apr 2010 05:24
"Peter Flass" <Peter_Flass(a)Yahoo.com> wrote in message news:hphsdl$tk7$2(a)news.eternal-september.org... > Morten Reistad wrote: >> In article <jfudndz3ZJOCTinWnZ2dnUVZ8uKdnZ2d(a)eclipse.net.uk>, >> Dave <g8mqw(a)yahoo.com> wrote: >>> "Patrick Scheible" <kkt(a)zipcon.net> wrote in message >>> news:w9zfx3fktul.fsf(a)zipcon.net... >> >>>> IBMs were leased. Would IBM continue to support a computer that had >>>> some academics' experimental hardware hooked up to it? Could new and >>>> experimental device drivers be added to IBM's OS? These might be as >>>> important as the machine's architecture. >>> Well post anti-trust they had too. And I know certainly in the UK, we >>> had lots of weird kit hooked into various University mainframes using a >>> wide variety of interfaces. The oldest I remember NUNET/NUMAC (I think) >>> used PDP/11s acting as IBM2708/3708 concentrators. There were "Browns >>> Boxes" for X.25 and some how Cambridge Ring got connected in at Leeds >>> but I think that was actually an Amdahl at that time... >> >> This was how mainframe networks were built. A cobbled-together mess >> of protocols, none from IBM. Philips had some gear that was very >> popular in banking/insurance networks in the mid 1970's, using >> datex (circuit switched 1200 bps) to access a remote network, and >> local end stations with local intelligence and some spooling. >> >> Similar stuff was used for travel agencies; but they used multidrop >> bisync; but still with some local intelligence. And the transaction >> screens became a nightmare of cryptical commands to get it all into >> one transaction. >> >> None of this had a chance to scale beyond the corporate "network" >> it was deployed in. I saw a merger between two banks up close, where >> they really tried to integrate. No such luck. They had to replace >> so they only had one network. Stats from internet exchange points still >> notice the encapsulated >> bisync, x25, frame relay etc. in the protocol fields. These legacy >> bits used 3% of the ix bandwidth as late as 2005. > > It's especially sad seeing the difference between the 360 architecture, > that was so well documented up-front, and SNA, that seeems to be whatever > could be made to work at the time. It's like two different companies. Not really, SNA is more software than hardware. So whilst S/360 and S/370 are well defined there are three very different mainstream Operating systems that run on that hardware. There is OS (aka MFT, MVT, MVS...), VM/CMS (Well really two OS's in one, CP and CMS), and DOS. All these operating systems have very different internal architectures so in general a program written for VM won't run on MVS or DOS, although the converse may be true as VM provides limited emulation of both MVS and DOS interfaces, and has a command to switch between the two. These all came from IBM. In addition to the three mainstream IBM distributed OS's there are many many more such as Music, MTS, TPF, TSS, ULTRIX, AIX/370 and Linux. Then if you look deeper into OS you find that there are multiple, incompatible TP environments (CICS and IMS) databses (IMS and DB2) etc etc etc. So whilst S/360 is consitant and well defined the software is far from it. Then when looking at SNA you have two things to remember. Firstly that it was late in coming to the scene and was designed to be a migratory way forward. Secondly it despite marketing messages, originally it only ran on OS(i.e. MVS), so other OS's continued with their own developments. So to me very much the same company. Lots of incompatible products. Often no proper roadmap only "Statements of Direction" (SODs) for guidance.... Dave |